lable at ScienceDirect

Journal of Archaeological Science 37 (2010) 569–576

Contents lists avai

Journal of Archaeological Science

journal homepage: ht tp: / /www.elsevier .com/locate/ jas

Macusani obsidian from southern Peru: A characterization of itselemental composition with a demonstration of its ancient use

Nathan Craig a,*, Robert J. Speakman b, Rachel S. Popelka-Filcoff c,d, Mark Aldenderfer e,Luis Flores Blanco f, Margaret Brown Vega a, Michael D. Glascock c, Charles Stanish g

a Department of Anthropology, The Pennsylvania State University, 409 Carpenter Building, University Park, PA 16802, USAb Smithsonian’s Museum Conservation Institute, 4210 Silver Hill Rd, Suitland, MD 20746, USAc Archaeometry Laboratory, Research Reactor Center, University of Missouri, 125 Chemistry, 601 South College Avenue, Columbia, MO 6521, USAd Department of Chemistry, University of Missouri, 125 Chemistry, 601 South College Avenue, Columbia, MO 6521, USAe School of Anthropology, University of Arizona, Emil W. Haury Building, P.O. Box 21003, USAf Co-Director del Proyecto Arqueologico Ramis, Puno, Perug Department of Anthropology, University of California at Los Angeles, 341 Haines Hall, Los Angeles, CA 90065-1553, USA

a r t i c l e i n f o

Article history:Received 19 March 2009Received in revised form4 October 2009Accepted 17 October 2009

Keywords:ObsidianPeruPXRFINAA

* Corresponding author. Tel.: þ1 312 714 4785.E-mail addresses: ncraig@psu.edu (N. Craig), speak

rachel.popelka-filcoff@nist.gov (R.S. Popelka-Filcoff),(M. Aldenderfer), lfloresb@correo.unmsm.edu.pe (L. Fedu (M.B. Vega), glascockm@missouri.edu (M.D. Glaedu (C. Stanish).

0305-4403/$ – see front matter � 2009 Elsevier Ltd.doi:10.1016/j.jas.2009.10.021

a b s t r a c t

Transparent obsidian artifacts have been reported for the northern Lake Titicaca Basin. Based oninstrumental neutron activation analysis (INAA) of these artifacts a distinct chemical group was iden-tified. Yet, the location of the source of transparent obsidian in the southern Andes remained unreportedin the archaeological literature. This paper reports on the chemical composition and geographic locationof a source of transparent obsidian from the Macusani region of Peru. Through the use of INAA andportable X-ray fluorescence (PXRF) we demonstrate that Macusani obsidian or macusanite comprises(at least) two chemical groups. One of these groups was used for making artifacts during the ArchaicPeriod. Artifacts made of this obsidian were found more than 120 km from the source and yet, one-thirdof the obsidian artifacts encountered at Macusani were from the non-local source of Chivay which is215 km to the southwest.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

In many parts of the ancient world, obsidian, because of itsworkability, was a valued commodity for making chipped stonetools. Since the terminal Pleistocene obsidian has been an impor-tant material in the southern Peruvian Andes (Burger et al., 2000;Glascock et al., 2007; Grosjean et al., 1997; Nunez et al., 2002). Inthis paper, we report the location and elemental composition ofa source of visually striking clear and transparent volcanic glassthat was used for making chipped stone tools since at least LateArchaic times.

Obsidian sourcing studies allow one to describe and model localand long-distance interaction through the exchange of portable andvaluable objects (Burger et al., 2000:268). In southern Peru and

manj@si.edu (R.J. Speakman),aldender@email.arizona.edu

lores Blanco), mybvega@psu.scock), stanish@anthro.ucla.

All rights reserved.

northern Bolivia, the elemental signatures and locations of nineanthropogenically exploited obsidian sources have been estab-lished: Alca (Cotahuasi) (Burger et al., 1998a), Chivay (Colca),Puzolana (Ayacucho) (Burger and Glascock, 2000; Tripcevich,2007), Jampatilla (Puquio) (Burger et al., 1998b), Lisahuacho(Andahuaylas A) (Glascock et al., 2007), Potreropampa (Anda-huaylas B) (Glascock et al., 2007), Aconcagua (Puno) (Aldenderfer,1999:383), Sora Sora (Bolivia) (Brooks et al., 1997:450), and Quis-pisisa (Ayacucho) (Burger and Glascock, 2000) (Fig. 1). Also repor-ted in the archaeological literature are the compositions andlocations of four obsidian sources that were not used for makingchipped stone tools (Glascock et al., 2007): Cerro Ticllago, Yanar-angra, Uyo Uyo, and Caylloma.

Of the 1696 obsidian artifacts from Peru that have been analyzedand assigned to a known source, 94% derived from the Chivay,Quispisisa, and Alca sources (Glascock et al., 2007:529–530). In thehighlands of Peru, south of the Department of Ayacucho, most ofthe analyzed obsidian artifacts derive from either the Chivay or Alcasources. Still, it is ‘‘clear that in this region several other types ofobsidian were utilized on a limited basis’’ (Burger et al., 2000:274).The identification of rare obsidians when large collections are

Fig. 1. Map of the Central Andes showing known obsidian sources and archaeological sites with obsidian artifacts that are mentioned in the text.

N. Craig et al. / Journal of Archaeological Science 37 (2010) 569–576570

screened ‘‘presents an interesting phenomenon which has receivedlittle attention in the literature’’ (Burger and Asaro, 1978:67). Theseanomalies may occur in ‘‘contact zones between the obsidian andthe adjacent non-igneous geologic formations,’’ they may comefrom ‘‘igneous formations with small patches or inclusions ofobsidian,’’ or they may be ‘‘long-distance imports from unsampledregions’’ (Burger and Asaro, 1978:67). Identifying the source loca-tions of rare obsidians is of anthropological interest because itpermits the definition of additional points of articulation inregional trade networks.

This paper details new observations regarding one type of rareand transparent obsidian from the southern Andes.

2. Examples of transparent obsidians in the archaeologicalrecord of the southern Andes

In the literature on archaeological obsidians in the south-centralAndes, more than five examples of artifacts produced from trans-parent obsidian have been reported. The analysis of five of suchartifacts revealed that they formed an identifiable chemical groupthat was termed Rare 9 Type obsidian (Burger et al., 2000).Although the Rare 9 Type obsidian was not recognized in theoriginal Lawrence Berkeley Laboratory instrumental neutron acti-vation analysis (INAA) study undertaken in the 1970s, it wasidentified in subsequent analyses (Burger and Asaro, 1977; Burgeret al., 2000:291). The material is reportedly ‘‘chemically similar to,but distinct from, Chivay Source obsidian’’ (Burger et al., 2000:296),and the geologic source of this material remains unknown.

The earliest reported archaeological instance of Rare Type 9obsidian occurs at the type-site of Qaluyu in the far northern Titicacaregion, during the Early Qaluyu phase (ca. 1300 BC) (Burger et al.,2000:289 Table III). The artifact is a triangular concave based

projectile point (Burger et al., 2000:290 Figure 295, specimen ‘‘d’’(292B/296–296, 8061-$)). In Karen Mohr-Chavez’s Unit C at Qaluyu,‘‘one additional transparent point (visually like Rare 9 Type)’’ wasreported. However, based on NAA it was suggested that the materialwas ‘‘apparently not obsidian’’ (Burger et al., 2000:298). AnotherRare Type 9 obsidian artifact was recovered from excavations ofEarly Horizon (ca. 1500–200 BC) contexts at Taraco, a major earlysettlement in the northern Lake Titicaca Basin. The artifact isa ‘‘[u]tilized flake with ground edge from a late Early Horizon level atTaraco (31B/36, 7013/)’’ (Burger et al., 2000:316–317, Figure 310artifact (h)). Rare Type 9 obsidian also has been reported from EarlyHorizon contexts at Pikicallepata and Cancha Cancha Asiruni (Burgeret al., 2000:310). ‘‘[Alfred] Kidder recovered colorless obsidian froma Late Chiripa level as well as from Early Tiahuanaco levels at Tia-huanaco (K. Chavez and Chavez, n.d.-a, b); it is tempting to speculatethat they could be Rare 9 Type, which is also colorless’’ (Burger et al.,2000:319). Thus with one possible exception from Tiwanaku, thetransparent Rare Type 9 obsidian is reported from the northernTiticaca Basin, and its use is limited to ca. 2000–200 BC.

In August of 2005, Speakman and Popelka-Filcoff used portableX-ray fluorescence (PXRF) to analyze all of the obsidian bifaces andprojectile points that had been recovered during pedestriansurveys of the Huenque (Klink, 1998, 2005), Ilave (Aldenderfer andde la Vega, 1996; Klink and Aldenderfer, 1996), lower Ramis(Stanish and Plourde, 2000), and Huancane (Stanish and Plourde,2000) drainages of the Lake Titicaca Basin (Craig et al., 2007). Thecollection includes two bifacial artifacts from the sites of HU-496and HU-113 in the Huancane region that were both made froma visually striking transparent natural glass. The specimen HU-496is a Late-Terminal Archaic stemmed projectile point that showsevidence of retouch. The piece from HU-113 is a broken andunfinished non-diagnostic biface.

N. Craig et al. / Journal of Archaeological Science 37 (2010) 569–576 571

Initial results of PXRF screening indicated that 1) the composi-tion of the two Huancane artifacts were similar and 2) they did notcorrespond to any of the south-central Andean sources with datarecorded at the University of Missouri Research Reactor (MURR).Because of the exceedingly high Rb concentrations (>1000 ppm)measured for these artifacts, we initially thought there wasa possibility that the artifacts may have originated from a north-west Argentine source. The reason for this was that at the time theonly obsidian known (to us) from the Andes to have Rb values inexcess of 600 ppm was from Argentina. Additionally, there is oneunknown Argentine obsidian source that has Rb values in excess of1000 ppm. If HU-496 and HU-113 were from Argentine obsidiansources, this would have been evidence of exceedingly long-distance trade. Consequently, more detailed INAA characterizationof the HU-496 and HU-113 specimens was undertaken and theresults confirmed that the two transparent obsidian artifacts didnot correspond with any obsidian that previously had beenanalyzed at MURR.

In 2007, we fortuitously identified a report that provided NAAdata for numerous obsidian sources, including one type from Perureferred to as Macusani (MacDonald et al., 1992). At the same time,we also discovered that samples of Macusani obsidian were housedin the rock and ore collections of the Department of MineralSciences at the Smithonsian’s National Museum of Natural History.Our subsequent investigation revealed that this glass had beenknown to geologists for more than eighty years, especially in thearea of fission track dating. The source has a complex geologichistory that involved multiple eruptions. We review this geologichistory because it helps to explain the chemical variability exhibi-ted by the obsidians from the source. Interestingly the source doesnot seem to be well known to archaeologists. Although Macusaniobsidian was mentioned in passing by Hughes (1998), to ourknowledge, the use of Macusani obsidian to make chipped stonetools had never been publisheddalthough the title of a recentpresentation by Louis Fortin (2006) would seem to indicate thatartifacts made from Macusani obsidian are known to a few arche-ologists working in Peru. In 2008, a visit was made to the MacusaniBasin to identify the exact location of Macusani obsidian depositsand to collect samples of the material for elemental characteriza-tion and comparison to the HU-496 and HU-113 specimens. Duringthis excursion to the Macusani area, numerous anthropogenically-unaltered obsidian pebbles and fourteen obsidian artifacts wereencountered (Fig. 2).

3. Research area and geology of the Macusani obsidian source

Having a mean elevation of 4400 m above sea level, the MacusaniBasin is a quadrilateral depression that is formed by the5000–6000 m peaks of the Cordillera Oriental (Cheilletz et al.,1992:309). The Macusani area is underlain by Paleozoic rocks. To thenorth, there are Carboniferous, Permian, and intrusive pre-creta-ceous rocks. To the south, there are upper cretaceous sedimentaryrocks (Barnes et al., 1970). Sillar flows are present throughout muchof the Macusani region. These flows are part of the extensive ash-flow deposits that mantle the slopes and high plain of the Andesmountains in Peru, Bolivia, Chile, and Argentina (Barnes et al.,1970:1540). The Macusani volcanics of the Meseta de QuenamariField largely consist of unwelded, crystal rich rhyolitic ash-flow tuffsthat contain a mixture of ash- and lapilli- sized pyroclastic fragmentsthat include pumice shards and lithic debris. The lithic debris largelyconsists of very low-grade or unmetamorphosed pelites, quartzites,and limestone (Cheilletz et al., 1992:310).

The first reports of Peruvian transparent natural glass to surfacein the academic literature described a material called ‘‘Pau-cartambo glass’’. Found as pebbles made of transparent natural

glass with a slight yellow-green tint with distinctive etching on thesurface, they were considered tektites (Linck, 1926). Based on theappearance, crystal inclusions, specific gravity, and refractive indexPaucartambo glass was found to be identical to Macusani obsidian.Given these similarities it is thought that what had initially beendescribed as ‘‘Paucartambo glass’’ was probably from the Macusaniarea and not from Cuzco (Martin and de Sitter-Koomans, 1955).Thus, ‘‘Paucartambo glass’’ (Linck, 1926) is likely Macusani obsidianor macusanite (which has also been referred to as ‘‘Macusani glass’’)(Pichavant et al., 1987:360).

It was not until the mid-1930s that natural glass from theMacusani region was first described, and in these early studies thematerial was attributed to an igneous rather than a celestial origin(Martin, 1934; Heide, 1936; Martin and de Sitter-Koomans, 1955).Though still in the mid-1960s studies were published in which itwas proposed that Macusani obsidian exhibited a composition thatwas ‘‘intermediate between that of sedimentary and igneousrocks’’; that ‘‘this Peruvian glass is unique’’, and it was ‘‘still notpossible to formulate an acceptable theory about its origin’’ (Elliottand Moss, 1965:424).

Macusani obsidians occur in a variety of colors which includetranslucent-green, opaque milky-green, translucent pale yellow, oropaque red-brown spots irregularly dispersed in a translucent-green matrix (Bigazzi et al., 1997; Pichavant et al., 1987). Somespecimens of macusanite are layered or flow banded (Barnes et al.,1970:1541). Opacity in the milky-green varieties is caused by thepresence of ‘‘abundant fluid inclusions of strongly irregular shape’’(Pichavant et al., 1987:360).

Macusanite pebbles are most frequently encountered alongseasonal streams in fluvio-glacial sediments on the southernborder of the Macusani ignimbrite field (Poupeau et al., 1993a:297).Previously, Macusani obsidian was first reported from Caluyo Mayo(Barnes et al., 1970), later at Chilcuno Chico (Arribas and Figueroa,1985; Valencia and Arroyo, 1985), and then at Samillia (Poupeauet al., 1993a). Though it had been established that macusanite wasnot a tektite, based on its presence in reworked fluvio-glacialsediments the origin of the material was still not well established.

In 1964, a Macusani obsidian pebble that had been recoveredfrom Caluyo Mayo was among the first materials ever dated bymeans of the fission track method; the initial age estimate was4.3� 0.4 Ma (Bigazzi et al., 2005:586; Fleischer and Price, 1964:758Table 753; Fleischer et al., 1965; Poupeau et al., 1993a:301). In thelate 1960s, a Macusani obsidian pebble from Caluyo Mayo anda biotite separate from the ash-flow tuffs were dated by the K–Armethod, and the results indicated ages of 4.2 � 1.5 and 4.1 � 1 Marespectively (Barnes et al., 1970:1543). Due to similarities in thechemical and mineralogical composition and the ages of thedeposits, it was concluded that the Macusani obsidian and the ash-flow tuffs probably originated during the same volcanic episode(Barnes et al., 1970:1545). 40Ar/39Ar dating indicates that theignimbrites of the Macusani Field erupted in three brief episodes:10� 1, 6.8–8, and 4�1 Ma (Cheilletz et al., 1990:343). The ash-flowtuffs are the youngest eruptive system of the Cordillera de Carabayasegment of the Central Andean Inner Arc (Cheilletz et al.,1990:341–342; Cheilletz et al., 1992:309).

Rb–Sr dating of two pebbles from Chilcuno Chico produced anage of 4.9 Ma (Pichavant et al., 1987:366). Rb–Sr dating of a mac-usanite pebble from Caluyo Mayo produced an age of 4.4 Ma(Pichavant et al., 1987:366). This result is in good agreement with FTand K–Ar ages from the same deposit (Barnes et al., 1970). In onestudy, fission-track plateau ages for Macusani obsidian samplescollected from Caluyo Mayo and Chilcuno Chico ranged from6.7� 0.3 to 7.0� 0.4 Ma (Cheilletz et al.,1992:311). In another study,FT plateau ages were determined for six Macusani obsidian pebblesfrom Chilcuno Grande. Five of the samples clustered around 7�1 Ma

Fig. 2. Map of the Macusani obsidian source showing the location of source samples and archaeological artifacts mentioned in the text.

N. Craig et al. / Journal of Archaeological Science 37 (2010) 569–576572

while the sixth sample produced an age of 4.76� 0.18 Ma (Poupeauet al., 1993b:502). Plateau ages for Macusani obsidian pebbles fromthe Caluyo Mayo area are 5.36–6.06 Ma (Poupeau et al., 1993b).These results are in agreement with prior FT age estimates of5.52�0.26 Ma and K–Ar estimates age estimates of 5.67�0.1 Ma forthese samples (Poupeau et al., 1992).

It has been suggested that Macusani obsidian should be used foran inter-laboratory fission track age calibration standard (Bales-trieri et al., 1996; Bigazzi, et al., 1997; McCorkell and Naeser, 1990;Miller and Wagner, 1981). However, the reference age is not wellconstrained. Fission-track plateau ages for Macusani obsidianpebbles range from 4.3 to 7.9 Ma (Miller et al., 1990; Miller andWagner, 1981; Naeser et al., 1980; Poupeau et al., 1993a,b, 1992).K–Ar ages from 4.2 � 1.5 to 5.67 Ma have been reported (Barneset al., 1970; Poupeau et al., 1992).

FT, K–Ar, and 40Ar/39Ar ages indicate that the Upper-Miocenevolcanic activity of the Macusani ignimbrite field was largelyconcentrated between 6.7 and 8 Ma, and this might have persisted toas late as 4.7 Ma with the volumetrically less important magmatismresponsible for the Macusani obsidian formation (Poupeau et al.,1993a:303). FT plateau ages indicate that the Macusani obsidianproducing volcanism occurred 4.8–6.9 Ma, possibly in the form ofdiscrete events (Poupeau et al., 1992:295). Macusani obsidians mayhave resulted from different volcanic eruptions (Poupeau et al.,1992), and this could explain the differences in ages (Osorio et al.,2003:411). Thus, though Macusani obsidian may have formedduring a single event that occurred around 5.5 � 1 Ma, it is alsopossible that multiple deposits formed in discrete events during thistime (Poupeau et al., 1993a:303). Slight variations in the chemical

compositions among different pebbles suggests that Macusaniobsidian may have formed in different eruptions (Poupeau et al.,1993a:303).

In 1988, inclusions of Macusani obsidian were found in ash-flowtuffs in the Chapi and Chilcuno Chico areas; these inclusions haveonly been observed in tuffs that belong to the upper cooling units(Pichavant et al., 1987:360). Examples of Macusani obsidian that areencountered as inclusions in the ash-flow tuffs are observed to beof the translucent-green variety, the surfaces are finely etched, andthe pebbles are generally of a smaller size (<2 cm) (Pichavant et al.,1987:360). In the Chapi area, obsidian clasts up to 5 cm in size werefound interbedded within the ash-flow tuffs (Cheilletz et al., 1992).From this observation, it was suggested that the Macusani obsid-ians found in fluvio-glacial sediments are either tephra ejecta thathave remobilized from the ash-flow tuffs or are remnants of flowdeposited glass located on the upper-most unit of the volcanicsequence that has subsequently eroded (Cheilletz et al., 1992:310).That the pebbles have a slightly flattened ovular shape and anetched surface, supports the model that Macusani obsidian frag-ments were erupted as clasts within the air-fall tephra. Followingthis, alluvial and glacial alterations separated the inclusions fromthe host tuff and resulted in the pebbles’ deposition in fluvio-glacialstream sediments (Pichavant et al., 1987:369). Flow depositedobsidian has not been reported in the Macusani region.

Based on Sr and O isotopes, Macusani obsidian can be dividedinto two groups. One group consists of samples obtained fromChilcuno Chico and Caluyo Mayo while the other group consists ofsamples from the Chapi inclusions. Each of these groups wasprobably formed by different magma (Pichavant et al., 1987:369).

Table 1Elemental concentrations of artifacts HU-113 and HU-496 analyzed by INAA atMURR and the macusanite sample analyzed by MacDonald et al. (1992).

Element HU-113 HU-496 Macusania

Al (%) 8.99 7.81 8.28K (%) 2.95 2.91 2.97Mn (%) 0.05 0.05 0.06Na (%) 3.10 2.80 3.10Bab <25 <25 <50Ceb 0.24 0.32 3Cs 524 529 503.0Dy 0.5 1.0 n.r.Eu <0.02 <0.02 <0.04Fe 3993 3965 3498Hf 1.4 1.7 1.3Lab 0.33 0.34 1Lu <0.04 <0.04 0.04Ndb <2.0 <2.1 <20Rb 1102 1111 1100Sb 4.5 4.6 3.9Sc 2.4 2.5 2.3Smb 0.5 0.6 1.5Sr <5 <5 1Ta 21.2 21.4 22.1Tb 0.1 0.2 0.17Th 1.9 1.7 1.3U 28.8 37.2 16.9Yb 0.3 0.5 0.3Zn 94.1 94.9 106.0Zrb <50 <50 28

a Values from MacDonald et al. (1992).b MURR values corrected for U-fission product interference (Glascock et al., 1986).

N. Craig et al. / Journal of Archaeological Science 37 (2010) 569–576 573

The Chilcuno Chico and Caluyo Mayo samples have higher 87Sr/86Srratios than either of the two inclusions from Chapi (Pichavant et al.,1987:365, Table 365).

4. Sample location and description

We investigated two areas within the Macusani Basin for thepresence of both natural obsidian deposits and artifacts. The firstlocation was about 11 km south of the town of Macusani, near thepueblo of Accoyo. Here a sparse lithic scatter that included bothclear transparent and black obsidian artifacts was encounteredalong about 500 m of the southern edge of a 7.0 ha bofedal (moor).The surface scatter included temporally diagnostic projectilepoints, bifaces, edge-modified flakes, cortical biface thinning flakes,and general debitage. The second study area was located about22 km northwest of the town of Macusani, along the active flood-plain of the Rio Isivilla. In this locality, numerous anthropogeni-cally-unaltered pebbles of clear pale-green transparent obsidianwere encountered. Several cortical flakes and a projectile pointmade out of a visually similar pale-green transparent material werealso found in the same context. A single biface thinning flake ofclear yellow-green material with red banding was recovered fromRio Isivilla (sample XRF ID006). In addition, a single black obsidianflake also was encountered in the floodplain of the Rio Isivilla.A clear slightly yellow-green color dominates the materialsencountered near Accoyo and in the sediments of the Rio Isivilla.Between the two sampling locations, a total of fourteen obsidianartifacts were encountered.

5. Methods of elemental analysis

As indicated above, samples from HU-496 and HU-113 wereanalyzed by INAA at MURR using standard analytical parameters(e.g., Glascock et al., 2007). All fourteen obsidian artifacts anda sample of 6 unmodified Macusani obsidian pebbles wereanalyzed at the Smithsonian’s Museum Conservation Institute witha Bruker Tracer III–V handheld XRF instrument. An additional6 macusanite specimens from the Smithsonian Institution’s rockand ore collection were also included in the analysis. XRF analysesof the obsidian artifacts permitted quantification of the followingelements: manganese (Mn), iron (Fe), rubidium (Rb), strontium(Sr), zirconium (Zr), and niobium (Nb). All obsidian samples wereanalyzed as unmodified samples. The Tracer III–V is equipped witha rhodium tube and a SiPIN detector with a resolution of ca. 170 eVFHWM for 5.9 keV X-rays (at 1000 counts per second) in an area of7 mm2. All analyses were conducted at 40 keV, 15 mA, usinga 0.076-mm copper filter and 0.0305 aluminum filter in the X-raypath for a 200-s live-time count. The spot size on this particularinstrument is ca. 4.0 mm diameter which facilitates analysis ofsmaller artifacts. Peak intensities for the above listed elementswere calculated as ratios to the Compton peak of rhodium, andconverted to parts-per-million (ppm) using linear regressionsderived from the analysis of 15 well characterized obsidian samplesthat previously had been analyzed by NAA and/or XRF.

6. Results of elemental analysis

The results of elemental characterization of the HU-496 andHU-113 samples as determined by means of INAA are reported inTable 1. Compositional data derived from the PXRF analyses of the6 anthropogenically-unmodified pebbles obtained from Peru forthis study, the six samples obtained from the Smithsonian’s rockand ore collection, and the fourteen artifacts are reported in Table 2.

Comparison of the INAA data generated for artifacts fromHU-496 and HU-113 with data presented in MacDonald et al. (1992)

confirms that these artifacts are indeed made from Macusaniobsidian (Table 1). XRF analyses also indicate that the sourcesamples we collected have a different chemical signature thanthose housed at the Smithsonian (Table 2). This was expected giventhat differences in fission track dates for Macusani obsidian, asdiscussed above, indicate multiple glass forming eruption events,and the possibility of different chemical signatures. The Smithso-nian Macusani obsidian samples also are cobble/fist sized whereasthe Rio Isivilla samples collected by us are approximately half thesize (or smaller). To denote the differences in composition, we arereferring to the two different chemical signatures as Macusani-1and Macusani-2, which correspond to the Smithsonian collectionthat come from Caluyo Mayo and to our Rio Isivilla samples,respectively. Of the fourteen obsidian artifacts from the MacusaniBasin that were analyzed, nine (64%) were crafted out of materialobtained from the Macusani-1 source and five (36%) were made outof material obtained from the Chivay source (Fig. 3). Given that theMacusani-1 obsidian occurs in a larger package size, it is not toosurprising that all of our Macusani obsidian artifacts have theCaluyo Mayo (e.g., Macusani-1) chemical signature.

7. Discussion

INAA and PXRF elemental analysis of artifacts and natural glasspebbles provide the first demonstration that prehispanic peoples ofthe south-central Andes used Macusani obsidian for making chip-ped stone tools. Prior studies suggested that Macusani obsidiancomprised at least two chemical groups, and it was proposed thatthis chemical variability likely represented glass deposits thatformed during different eruptive events. Our results confirm thepresence of two chemical groups at the Macusani obsidian source.Whatever geologic processes caused this compositional heteroge-neity, we were able to document that material from at least one ofthese chemical groups was used for making chipped stone artifacts.Given the nature of this geologic deposit, the identification of

Table 2Elemental concentrations of source rocks and artifacts from the Macusani region analyzed by PXRF.

Lab ID Chem. group Sample type Mn Fe Rb Sr Zr Nb

SI-2143-20 Macusani-1 Source 566 4284 1138 5 10 36SI-2143-37 Macusani-1 Source 397 4430 1172 5 7 37SI-2143-27 Macusani-1 Source 421 4367 1175 6 5 35SI-2143-25 Macusani-1 Source 465 4398 1205 4 9 39SI-2143-23 Macusani-1 Source 475 4166 1173 5 11 37SI-2143-12 Macusani-1 Source 514 4448 1198 5 8 37

Mean 473 4349 1177 5 8 37Std Dev 56 97 22 1 2 1

MAC22-1 Macusani-2 Source 646 3276 1349 8 4 34MAC22-2 Macusani-2 Source 657 3262 1356 8 5 34MAC22-3 Macusani-2 Source 532 3287 1315 6 4 28MAC22-4 Macusani-2 Source 665 3260 1354 9 12 36MAC23-1 Macusani-2 Source 654 3190 1366 8 5 36MAC23-2 Macusani-2 Source 530 3418 1403 7 8 36

Mean 614 3282 1357 8 6 34Std Dev 59 68 26 1 3 3

XRF001 Chivay Flake 712 4979 241 40 66 16XRF002 Chivay Flake 616 4737 226 35 69 13XRF003 Chivay Edge-modified Flake 568 4993 238 47 66 15XRF004 Chivay Serrated 5C Projectile Point 593 4940 245 39 68 15XRF005 Chivay Flake 499 4776 234 40 64 15XRF006 Macusani-1 Biface Thinning Flake 291 4381 1229 4 10 33XRF007 Macusani-1 Edge-modified Cortical Flake 351 4166 1150 5 4 32XRF008 Macusani-1 Cortical Flake 351 4129 1135 3 7 33XRF009 Macusani-1 Biface Fragment 283 4076 1128 2 2 33XRF010 Macusani-1 Flake 438 4234 1168 3 13 33XRF011 Macusani-1 Serrated 4D Projectile Point 367 3999 1139 4 13 34XRF012 Macusani-1 Serrated 4H Projectile Point 392 4153 1155 5 5 36XRF013 Macusani-1 Edge-modified Cortical Flake 450 4136 1115 5 6 33XRF014 Macusani-1 Cortical Biface Thinning Flake 312 4715 1136 0 2 26

Fig. 3. Rubidium–strontium plot of source rocks and artifacts from the Macusani region analyzed by PXRF. The ellipses surrounding each group are drawn at the 95% confidencelevel. Confidence ellipses for Alca, Chivay, and Quispisisa are calculated using data published in Glascock et al. (2007). Inset: Iron–Manganese plot showing additional separation ofthe geologic samples assigned to the Macusani-1 and Macusani-2 chemical groups.

N. Craig et al. / Journal of Archaeological Science 37 (2010) 569–576574

N. Craig et al. / Journal of Archaeological Science 37 (2010) 569–576 575

multiple chemical groups of clear obsidian in archaeologicalassemblages from the south-central Andes can be expected.

The other nine anthropogenically exploited sources (Quispisisa,Jampatilla, Lisahuacho, Potreropampa, Puzolana, Alca, Chivay,Aconcagua, and Sora Sora) and the four known geologic sourceswhich were not exploited for tool raw materials (Yanarangra, UyoUyo, Caylloma, and Cerro Ticllago) are all located on the westernside of the Andes in an ‘‘arching line between 3000 and 5000 mabove sea level’’ (Tripcevich, 2007:181, Figure 183–185, 331,Figure 184–112). Based on the present evidence, Macusani obsidianis the only source located on the eastern cordillera that wasexploited for producing chipped stone tools.

The discovery of three temporally diagnostic Archaic projectilepoint forms (HU-496, XRF011, and XRF012) crafted out of Macusaniobsidian indicates that exploitation of this source dates to pre-ceramic times. One of these points was encountered in the LateTiticaca Basin (HU-496), and it indicates that Macusani obsidianwas transported long distances (>120 km) from the source duringthe Archaic. Synthesis of prior research indicated that evidence forthe use of Rare Type 9 obsidian chemical group appears first duringEarly Qaluyu (ca. 1300 BC) (Burger et al., 2000:289 Table III). If theRare Type 9 group corresponds to Macusani obsidian, then thismaterial was used from the Archaic to the Formative Period. Burgeret al. (2000) state that even Rare 9 Type obsidian, which is alwaysvisually distinct because of its transparency, looks like another typethat is chemically similar to, but distinct from, Chivay obsidian.Given the unusually high Rb concentration for Macusani obsidian, itis clear that Macusani obsidian is not chemically similar to Chivayobsidian. This raises the question of whether a second geologically/geographically discrete source of colorless obsidian exists in Peru.Unfortunately, we will not know the answer to this question untilsuch time that we have the opportunity to analyze/reanalyze someof the translucent obsidian artifacts that have been reported in theliterature.

Three obsidian projectile points were found in the MacusaniBasin. XRF011 and XRF012 are stemmed forms that are morpho-logically and temporally diagnostic to the Late Archaic. XRF012exhibits an expanding stem. This type has a known distributionthat is restricted to the Cuzco Basin (Bauer, 2007; Bauer et al., 2007;Klink, 2007) and the northern and eastern parts of the Lake TiticacaBasin. Despite systematic pedestrian survey, this expanding stem-med form has not been encountered in either the Rio Ilave(Aldenderfer and de la Vega, 1996; Klink and Aldenderfer, 1996) orRio Huenque (Klink, 1998, 2005) drainages. XRF004 is a triangularconcave base form. This type spans the Archaic–Formative transi-tion (ca. 2000–1500 BC), and in the Andean highlands it exhibitsa very cosmopolitan distribution.

Intriguingly, all three of the projectile points found at theMacusani source were serrated. HU-496 is heavily retouched, and itis not possible to determine if it was serrated at one point. HU-113is not serrated, but it is also a simple biface and is not a well-formedprojectile point. It has been noted that in the Rio Ilave drainage,projectile points made out of obsidian are serrated or denticulatedsignificantly more often than projectile points made out of othermaterials (Craig, 2005:683–688). The fact that three of the fivemacusanite bifaces we analyzed are serrated is consistent with thispreviously noted trend.

Even at the Macusani source, one-third of all the obsidian arti-facts that we encountered were crafted out of material that derivedfrom the Chivay source. Chivay is located 215 km to the southwest,across the altiplano, on the western cordillera. Thus, non-localobsidian was desirable to those inhabiting the Macusani region.This observation reinforces the argument that in the south-centralAndean highlands, obsidian was a chipped stone raw material thatwas desired for largely social reasons (Aldenderfer, 2005; Craig,

2005; Frye et al., 1998). The Macusani Basin must have beenarticulated with exchange networks that crossed the altiplano andlinked to the Chivay source. The Macusani Basin is the farthesttoward the Ceja de Selva or eastern slopes of the Andes in which thepresence of Chivay obsidian has thus far been documented. Weenvision three scenarios to explain the presence of Chivay obsidianin the Macusani Basin: 1) people from Chivay traveled directly toMacusani; 2) people from Macusani traveled to Chivay; and/or 3)people from Macusani exchanged with third parties who hadaccess to sufficient quantities of Chivay obsidian that they werewilling to trade it away. Further work is required to evaluate thesemodels of exchange.

8. Concluding remarks

It has been suggested that the visual attractiveness and physicalproperties that yield sharp-edged, conchoidal fractures would haveadded to the appeal of a given obsidian source material (Glascocket al., 2007:522). Macusani obsidian is visually striking; it is rela-tively free of phenocrysts and other inclusions. Thus, based onthese criteria one would expect that Macusani obsidian would havebeen highly desirable. Macusani obsidian only occurs as smallersized clasts in ash-flow tuff or in fluvio-glacial stream beds. OnlyMacusani-1 obsidian, which occurs in larger sized clasts thanMacusani-2, was used for making chipped stone artifacts. Thougha potentially desirable source of material, the small package size ofraw materials from the Macusani obsidian deposit would haverendered difficult the kind of intensive exploitation that has beenobserved at major obsidian sources, such as Chivay (Tripcevich,2007:583–584, 687–743). Still, the presence of multiple Macusaniobsidian artifacts in the northern LTB indicates that the materialwas transported more than 120 km away from the source. Thisserves as an indication that Macusani obsidian was articulated withlong-distance trade networks and was probably a highly desiredmaterial which was only available in relatively small quantities.

Acknowledgements

The authors would like to thank Albino Pilco Quispe (A.K.A. TheMcGyver of the Andes) for his keen eyes in the field and for hisabsolutely amazing ability to repair machines with little more thana Leatherman�. Special thanks are extended to Leslie Hale andLuElla Parks for bringing MacDonald et al’s. (1992) publication toour attention and for facilitating the loan of the Macusani obsidianhoused in the Smithsonian Institution’s National Museum ofNatural History. Speakman would like to acknowledge Kathy Tigheand Bruce Kaiser for their support and assistance with the XRFportions of this project. We gratefully acknowledge the support ofthe National Science Foundation grant BCS-0318500 which wasawarded to Aldenderfer to survey the upper Rio Ramis; grants BCS9905138, BCS-0003229, and BCS 0210158 which were awarded toStanish to survey the Huancane and lower Ramis; grant BCS-0405042 which was awarded to Glascock to purchase the PXRFinstrument; and generous support from the Curtis T. and Mary G.Brennan Foundation which was awarded to Craig to supportaspects of the field work.

References

Aldenderfer, M.S., de la Vega, E., 1996. Informe Final: Reconocimiento Arqueologicode la Cuenca del Rıo Ilave, Sub-region de Puno, Region ‘‘Jose Carlos Mariategui’’.Report submitted to Instituto Nacional de Cultura, Lima.

Aldenderfer, M.S., 1999. Cronologıa y Conexiones: Evidencias Preceramicas deAsana. In: Kaulicke, P. (Ed.), Boletin de Arqueologica PUCP. Pontifica UniversidadCatolica del Peru, Lima, pp. 375–392.

N. Craig et al. / Journal of Archaeological Science 37 (2010) 569–576576

Aldenderfer, M.S., 2005. Preludes to power in the highland late preceramic period.In: Vaughn, K.J., Ogburn, D.E., Conlee, C.A. (Eds.), Foundations of Power in thePrehispanic Andes. American Anthropological Association, Arlington, Virginia,pp. 13–35.

Arribas, A., Figueroa, E., 1985. Geologia y metalogenia de las mineralizacionesuraniferas de Macusani, Puno (Peru), Uranium Deposits in Volcanic Rocks. IAEA,Vienna. 237–254.

Balestrieri, M., Bigazzi, G., Oddone, M., 1996. The thermal column: a new irradiationposition for fission-track dating in the University of Pavia Triga Mark II nuclearreactor. Journal of Radioanalytical and Nuclear Chemistry 213, 99–108.

Barnes, V.E., Edwards, G., McLaughlin, W.A., Friedman, I., Joensuu, O., 1970. Mac-usanite occurrence, age, and composition, Macusani, Peru. Geological Society ofAmerica Bulletin 81, 1539–1546.

Bauer, B. (Ed.), 2007. Kasapata and the Archaic Period of the Cuzco Valley. CotsenInstitute of Archaeology at University of California Los Angeles, Los Angeles.

Bauer, B.S., Jones, B., Klink, C., Sutter, R.C., DeFrance, S.D., Burger, R.L., 2007. TheArchaic period of the Cuzco Valley. In: Bauer, B. (Ed.), Kasapata and the ArchaicPeriod of the Cuzco Valley. Cotsen Institute of Archaeology at University ofCalifornia Los Angeles, Los Angeles, pp. 122–124.

Bigazzi, G., Coltelli, Hadler Neto, J.C., Osorio Araya A.M., 1997. Provenance studies ofobsidian artefacts using the fission-track analysis in South America: an over-view. In: 49th Congreso Internacional del Americanistas, Quito, Ecuador.

Bigazzi, G., Hadler Neto, J.C., Iunes, P.J., Osorio Araya, A.M., 2005. Fission-trackdating of South American natural glasses: an overview. Radiation Measure-ments 39, 585–594.

Brooks, S.O., Glascock, M.D., Giesso, M., 1997. Source of volcanic glass for ancientAndean tools. Nature 386, 449–450.

Burger, R.L., Asaro, F., 1977. Trace Element Analysis of Obsidian Artifacts from theAndes: New Perspectives on Pre-Hispanic Economic Interaction in Peru andBolivia. Lawrence Berkeley Laboratory, University of California at Berkeley,Berkeley, pp. 1–93.

Burger, R.L., Asaro, F., 1978. Obsidian Distribution and Provenience in the CentralHighlands and Coast of Peru during the Preceramic Period. In: Contributions ofthe University of California Archaeology Research Facility, vol. 36, pp. 61–83.

Burger, R.L., Asaro, F., Tarwick, P., Stross, F., 1998a. The Alca obsidian source: theorigin of raw material for Cuzco type obsidian artifacts. Andean Past 5, 185–202.

Burger, R.L., Schreiber, K.J., Glascock, M.D., Ccencho, J., 1998b. The Jampatillaobsidian source: identifying the geological source of the Pampas type obsidianartifacts from southern Peru. Andean Past 5, 225–239.

Burger, R.L., Chavez, K.L., Chavez, S., 2000. Through the glass darkly: prehispanicobsidian procurement and exchange in southern Peru and northern Bolivia.Journal of World Prehistory 14, 267–362.

Burger, R.L., Glascock, M.D., 2000. Locating the Quispisisa obsidian source in theDepartment of Ayacucho, Peru. Latin American Antiquity 11, 258–268.

Cheilletz, A., Clark, A.H., Farrar, E., Arroyo Pauca, G., MacArthur, J.D., Pichavant, M.,1990. Stratigraphy and geochronology of the Macusani ignimbrite field: chro-nometer of the mio-pliocene geodynamic evolution of the Andes of SE Peru. In:Symposium International ‘‘Geodynamique Andine’’, Paris, pp. 341–344.

Cheilletz, A., Clark, A.H., Farrar, E., Pauca, G.A., Pichavant, M., Sandeman, H.A., 1992.Volcano-stratigraphy and 40Ar/39Ar geochronology of the Macusani ignimbritefield: monitor of the Miocene geodynamic evolution of the Andes of southeastPeru. Tectonophysics 205, 307–327.

Craig, N., 2005. The Formation of Early Settled Villages and the Emergence ofLeadership: a Test of Three Theoretical Models in the Rio Ilave, Lake TiticacaBasin, Southern Peru. Unpublished Ph.D. dissertation, Department of Anthro-pology, University of California at Santa Barbara, Santa Barbara, p. 995.

Craig, N., Speakman, R.J., Popelka-Filcoff, R.S., Glascock, M.D., Robertson, J.D.,Shackley, M.S., Aldenderfer, M.S., 2007. Comparison of XRF and PXRF for anal-ysis of archaeological obsidian from southern Peru. Journal of ArchaeologicalScience 34, 2012–2024.

Elliott, J., Moss, A., 1965. Natural glass from Macusani, Peru. Mineralogical Magazine35, 423–424.

Fleischer, R.L., Price, P.B., 1964. Fission track evidence for the simultaneous origin oftektites and other natural glasses. Geochimica et Cosmochimica Acta 28, 755–756. IN751–IN755, 757–760.

Fleischer, R.L., Price, P.B., Walker, R.M., 1965. Tracks of charged particles in solids.Science 149, 383–393.

Fortin, L., 2006. The archaeological importance of obsidian and Macusanite in theSouthern Peruvian Highlands. In: 14th Annual J. Louis Agassiz Symposium,Orno, Maine.

Frye, K.L., Aldenderfer, M.S., Glascock, M.D., 1998. The Aconcahua obsidian sourceand its Relation to south central Andean exchange systems. In: 38th AnnualAndean Meeting, Berkeley, CA.

Glascock, M.D., Nabelek, P.I., Weinrich, D.D., Coyeney Jr., R.M., 1986. Correcting foruranium fission in instrumental neutron activation analysis of high-uraniumrocks. Journal of Radioanalytical and Nuclear Chemistry 99, 121–131.

Glascock, M.D., Speakman, R.J., Burger, R.L., 2007. Sources of archaeological obsidianin Peru: descriptions and geochemistry. In: Glascock, M.D., Speakman, R.J.,Popelka-Filcoff, R.S. (Eds.), Archaeological Chemistry: Analytical Techniques andArchaeological Interpretation. American Chemical Society, Washington, DC, pp.522–552.

Grosjean, M., Nunez, L., Cartajena, I., Messerli, B., 1997. Mid-Holocene climate andculture change in the Atacama Desert, northern Chile. Quaternary Research 48,239–246.

Heide, F., 1936. Neue kristallfuhrende Glaser von Macusani in Peru, Natur-wissenschaften 24, pp. 281–282.

Hughes, R.E., 1998. On reliability, validity, and scale in obsidian sourcingresearch. In: Ramenofsky, A.F., Steffen, A. (Eds.), Unit Issues in Archaeology:Measuring Time, Space, and Material. University of Utah Press, Salt lake City,pp. 103–114.

Klink, C., Aldenderfer, M.S., 1996. Archaic period settlement patterns on the Peru-vian Altiplano: Comparison of two recent surveys in the southwestern LakeTiticaca Basin. In: Paper Presented at the 24th Annual Midwest Conference ofAndean and Amazonian Archaeology and Ethnohistory, Beloit.

Klink, C., 1998. Informe Final: Proyecto Arcaico del Altiplano 1997, ProspeccionArqueologica en el Valle del Rıo Huenque. Report submitted to InstitutoNacional de Cultura, Lima.

Klink, C., 2005. Archaic period research in the Rio Huenque valley, Peru. In:Stanish, C., Cohen, A.B., Aldenderfer, M.S. (Eds.), Advances in Titicaca BasinArchaeology. Cotsen Institute of Archaeology, Los Angeles, pp. 13–24.

Klink, C., 2007. The lithic assemblage at Kasapata. In: Bauer, B. (Ed.), Kasapata andthe Archaic Period of the Cuzco Valley. Cotsen Institute of Archaeology atUniversity of California Los Angeles, Los Angeles, pp. 31–77.

Linck, G., 1926. Ein Neuer Kristallfuhrender Tektit von Pucartambo in Peru. Chemieder Erde 2, 157–174.

MacDonald, R., Smith, R.L., Thomas, J.E., 1992. Chemistry of the Subalkalic SilicicObsidians. U.S. Government Printing Office, Washington, DC.

Martin, W.R., 1934. Are the ‘‘Americanites’’ Tektites?, Leidsche Geologische Mede-deelingen 6, pp. 123–132.

Martin, R., de Sitter-Koomans, C., 1955. Pseudotectities from Colombia and Peru.Leidse Geologische Mededelingen 20, 151–164.

McCorkell, R.H., Naeser, C.W., 1990. Natural glass interlaboratory reference mate-rials for fission track dating. International Journal of Radiation Applications andInstrumentation. Part D. Nuclear Tracks and Radiation Measurements 17, 422.

Miller, D.S., Wagner, G.A., 1981. Fission-track ages applied to obsidian artifacts fromSouth America using the plateau-annealing and the track-size age-correctiontechniques. Nuclear Tracks 5, 147–155.

Miller, D.S., Ebny, N., McCorkell, R.H., Rosenberg, P.E., Suzuki, M., 1990. Results ofinterlaboratory comparison of fission track ages for 1988 fission track work-shop. Nuclear Tracks and Radiation Measurement 17, 237–245.

Naeser, C.W., Izett, G.A., Obradovitch, J.D., 1980. Fission-track and K–Ar ages ofnatural glasses. Geological Survey Bulletin 1489, 1–31.

Nunez, L., Grosjean, M., Cartajena, I., 2002. Human occupations and climate changein the Puna de Atacama, Chile. Science 298, 821–824.

Osorio, A.M., Iunes, P.J., Bigazzi, G., Hadler N, J.C., Laurenzi, M.A., Norelli, P., Tello S, C.A.,Guedes, S., Paulo, S.R., 2003. Fission-track dating of Macusanite glasses withplateau and size correction methods. Radiation Measurements 36, 407–412.

Pichavant, M., Herrera, J.V., Boulmier, S., Briqueu, L., Joron, J.-L., Juteau, M., Marin, L.,Michard, A., Sheppard, S.M.F., Treuil, M., Vernet, M., 1987. The Macusani glasses,SE Peru: evidence of chemical fractionation in peraluminous magmas. In:Mysen, B.O. (Ed.), Magmatic Processes: Physicochemical Principles a Volume inHonor of Hatten S. Yoder, Jr. The Geochemical Society, Calgary, pp. 359–373.

Poupeau, G., Sabil, N., Villa, I.M., Bigazzi, G., Vatin-Perignon, N., Flores, P., Pereyra, P.,Salas, G., Arroyo, G., 1992. Fission-track and K–Ar ages of ‘‘macusanite’’ obsidianglasses, (SE Peru): geodynamic implications. Tectonophysics 205, 295–305.

Poupeau, G., Labrin, E., Sabil, N., Bigazzi, G., Arroyo, G., Vatin-Perignon, N., 1993a.Fission-track dating of 15 macusanite glass pebbles from the Macusani volcanicfield (SE Peru). International Journal of Radiation Applications and Instru-mentation. Part D. Nuclear Tracks and Radiation Measurements 21, 499–506.

Poupeau, G., Labrin, E., Sabil, N., Vatin-Perignon, N., Arroyo, G., Bigazzi, G., 1993b.Fission track dating of 18 macusanite glass pebbles. International Journal ofRadiation Applications and Instrumentation. Part D. Nuclear Tracks and Radi-ation Measurements 21, 583.

Stanish, C., Plourde, A., 2000. Informe Final: Archaeological Survey in the Huan-cane-Putina Valley, Peru. Report submitted to Instituto Nacional de Cultura,Lima.

Tripcevich, N., 2007. Quarries, Caravans, and Routes to Complexity: PrehispanicObsidian in the South-Central Andes. Unpublished Ph.D. dissertation, Depart-ment of Anthropology, University of California, Santa Barbara, Santa Barbara.

Valencia, J., Arroyo, G., 1985. Consideraciones Geoquimicas de los Indicios Urani-feros de Macusani, Puno (Peru), Uranium Deposits in Volcanic Rocks. IAEA,Vienna, pp. 275–288.